debdu

commit ce9e99275dc5d1494dd665eb31dfe7550c55f9b8
Author: Antoine Amarilli <a3nm@a3nm.net>
Date:   Fri, 25 Jul 2014 01:17:15 +0200

initial

Diffstat:
debdu.py
 | 
76
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

1 file changed, 76 insertions(+), 0 deletions(-)
diff --git a/debdu.py b/debdu.py
@@ -0,0 +1,76 @@
+#!/usr/bin/python
+
+import apt_pkg
+
+apt_pkg.init()
+cache = apt_pkg.Cache(None)
+depcache = apt_pkg.DepCache(cache)
+
+graph = {}
+roots = []
+results = []
+
+def init(p):
+  if p not in graph.keys():
+    graph[p] = {}
+    graph[p]['n_in'] = 0
+
+def dfs(v):
+  if graph[v].has_key('full'):
+    return graph[v]['full']
+  graph[v]['full'] = 0 # TODO this marking is wrong for cycles, think about it
+  s = graph[v]['size']
+  expl = [(s, 'the package itself', True, 1)]
+  for d in graph[v]['deps']:
+    #print "%s -> %s" % (v, d)
+    dfs(d)
+    if graph[d]['auto']:
+      s += graph[d]['full'] / graph[d]['n_in']
+      expl.append((graph[d]['full']/graph[d]['n_in'], d,
+        graph[d]['auto'], graph[d]['n_in']))
+    else:
+      expl.append((0, d, graph[d]['auto'], 1))
+  graph[v]['full'] = s
+  expl = sorted(expl, reverse=True)
+  if not graph[v]['auto']:
+    results.append((s, v, expl))
+
+for package in cache.packages:
+  if package.current_state != apt_pkg.CURSTATE_INSTALLED:
+    continue
+  version = package.current_ver
+  name = package.name
+  #print name
+  auto = depcache.is_auto_installed(package)
+  init(name)
+  graph[name]['auto'] = auto
+  if not auto:
+    roots.append(name)
+  graph[name]['size'] = version.installed_size
+  graph[name]['deps'] = set()
+  if 'Depends' not in version.depends_list_str.keys():
+    continue
+  for dep in version.depends_list_str['Depends']:
+    # TODO of course this is wrong...
+    for or_dep in dep:
+      try:
+        if cache[or_dep[0]].current_state != apt_pkg.CURSTATE_INSTALLED:
+          continue
+      except KeyError:
+        continue # Wrong wrong wrong!
+      graph[name]['deps'].add(or_dep[0])
+      init(or_dep[0])
+      graph[or_dep[0]]['n_in'] += 1
+
+for root in roots:
+  dfs(root)
+
+for result in sorted(results, reverse=True):
+  print str(result[0]) + ' ' + result[1]
+  for e in result[2]:
+    # TODO shared should be between non-descendant packages
+    print '  %d from %s%s' % (e[0], e[1],
+        (" shared between "+str(e[3])+" package(s)" if e[3] != 1 else '') if
+        e[2] else ' manually installed and accounted in a separate entry')
+  print
+